Note Demonstration of both Dev and Ops stages of the MLOps pipeline. The training script (including hyperparameter tuning) is present in the file train.py. The file app.py implements a gradio app for inference using the freshly trained model. Logs are synced every two minutes to a HuggingFace dataset.

Note Dashboard that implements Data Drift and Model Drift. The dashboard refreshes every 5 seconds.

Note Logs collected from the API and UI for machine failure. These logs are synced from the deployed model every 2 minutes.

Note An interface to execute drift checks manually. Useful for understanding the edge-cases where machine failure might occur.